In Silico Design and in Vitro Characterization of Universal Tyrosine

Communication Cite This: Biochemistry XXXX, XXX, XXX−XXX

pubs.acs.org/biochemistry

In Silico Design and in Vitro Characterization of Universal Tyrosine Kinase Peptide Substrates Laura J. Marholz,† Nicholas A. Zeringo,‡ Hua Jane Lou,‡ Benjamin E. Turk,*,‡ and Laurie L. Parker*,† †

Department of Biochemistry, Molecular Biology and Biophysics, University of Minnesota, 420 Washington Avenue Southeast, Minneapolis, Minnesota 55455, United States ‡ Department of Pharmacology, Yale University School of Medicine, P.O. Box 208066, 333 Cedar Street, New Haven, Connecticut 06520, United States S Supporting Information *

homogeneous sequence (e.g., EEEEYEEEE) or more commonly as a heterogeneous polymeric material. These substrates are commercially available and often included in PTK assay kits and have been used for decades in widespread applications to evaluate activity and to serve as pseudouniversal substrates in tyrosine kinase assays.4−8 Because PTKs often have strict substrate specificity, random mixtures of polyGlu-Tyr peptides are usually phosphorylated inefficiently, often to low stoichiometry.9 Because of the heterogeneous nature of the polymer, the same sample may contain components that a given kinase will phosphorylate but may also harbor sequences that act as inhibitors of enzymatic activity. This inconsistency impedes accurate comparison of activity between preparations of substrates and across different assay conditions. Also, these heterogeneous polymers are not compatible with highthroughput assays commonly employed for inhibitor screening (e.g., LanthaScreen or AlphaScreen technologies and other TRFRET applications), making it very challenging to develop new, efficient assays to target understudied kinases. In addition, comparing the relative activity of the same or different kinases from multiple preparations or assays is a poorly recognized confounding challenge in drug screening. Efficient, wellcharacterized, common substrates would enable assay standardization by calibrating specific activity between batches of enzyme and assay replicates. Among other benefits, having a standardized screening format would facilitate kinomewide profiling of inhibitor specificity, which is essential for assessment of potential off-target activity. Several in silico methods have been reported that attempt to predict the substrate specificity of kinases.10−13 Previous work in our laboratory developed the KINATEST-ID platform for kinase artificial substrate discovery.14 This pipeline uses an input data set of established literature-curated kinase substrate and nonsubstrate sequences to produce a position-specific scoring matrix (PSSM). The PSSM provides a comprehensive assessment of the amino acid preferences for a given kinase at multiple positions flanking the tyrosine phosphosite. An in silico library of candidate biosensor sequences can be generated and evaluated for predicted kinase specificity based on PSSMs derived from multiple kinases. In the work reported here, the KINATEST-ID workflow was used in a reversed fashion with

ABSTRACT: A majority of the 90 human protein tyrosine kinases (PTKs) are understudied “orphan” enzymes with few or no known substrates. Designing experiments aimed at assaying the catalytic activity of these PTKs has been a long-running problem. In the past, researchers have used polypeptides with a randomized 4:1 molar ratio of glutamic acid to tyrosine as general PTK substrates. However, these substrates are inefficient and perform poorly for many applications. In this work, we apply the KINATEST-ID pipeline for artificial kinase substrate discovery to design a set of candidate “universal” PTK peptide substrate sequences. We identified two unique peptide sequences from this set that had robust activity with a panel of 15 PTKs tested in an initial screen. Kinetic characterization with seven receptor and nonreceptor PTKs confirmed these peptides to be efficient and general PTK substrates. The broad scope of these artificial substrates demonstrates that they should be useful as tools for probing understudied PTK activity.

P

rotein tyrosine kinases (PTKs) are responsible for phosphorylating specific tyrosine residues in substrate proteins, which can regulate a variety of cellular signaling pathways. This post-translational covalent modification can affect the activity, subcellular localization, or stability of proteins, thereby modulating critical processes such as cell growth and proliferation, metabolism, differentiation, and migration.1 Because deregulation of these processes is a key event in cancer progression, PTKs have emerged as important drug targets, with many PTK inhibitors in current clinical use as cancer therapeutics.2 One of the key ways to identify small molecule PTK inhibitors is through activity-based screens that assay phosphorylation of a substrate in the presence of the compound. For well-studied kinases, efficient peptide or protein substrates are known for use in these assays. However, the availability of assays for the “orphan” kinome is less straightforward. Of the 90 receptor and nonreceptor tyrosine kinases encoded in the human genome,3 up to 90% are relatively understudied with few high-quality substrates identified. In the absence of an established substrate, a common approach is to use sequences containing tyrosine and various molar ratios of glutamic acid residues (polyGlu-Tyr), either as a © XXXX American Chemical Society

Received: January 12, 2018 Revised: February 22, 2018

A

DOI: 10.1021/acs.biochem.8b00044 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry

for a particular kinase. The usual workflow for this process would be to filter the resulting list to remove sequences that do not have predicted specificity to a kinase of interest. However, our goal was to identify sequences with the broadest possible activity, so we ranked sequences according to highest total summed score for all screened kinases. A summary of the two scored and ranked libraries is included in Tables S15 and S16. With these curated libraries in hand, we next sought to select a set of sequences for synthesis and empirical testing. The Screener tool removed the vast majority of the initial sequences from consideration but left several thousand high-scoring sequences from which to choose. In the end, a total of nine different peptide sequences were chosen to form a pilot library for testing in a low-throughput kinase panel activity screen. These peptides do not simply represent the nine highestscoring sequences that were output from Screener but rather were selected on the basis of a variety of factors, including considerations of additional details about known kinase preferences and particular chemical properties at certain positions (as discussed in our previous work14,15), to introduce a measure of diversity into this pilot library. The rationale for the sequence selection, synthetic procedures, and characterization data for these peptides are included as Supporting Information. The final list of peptides, their sequences, and their summed score from Screener are included in Table 1.

the goal of designing a substrate that can be phosphorylated by as many PTKs as possible. A truly universal PTK substrate based on the actual substrate sequence preferences of a panel of PTKs would enable efficient activity assays to be designed, even for tyrosine kinases for which no substrates are currently known, and used in many applications from further study of orphan kinase biology to drug or tool compound discovery. To identify a potential universal motif, the amino acid preference profiles for ABL, ARG, BTK, CSK, FES, FYN, HCK, JAK2, LCK, LYN, PYK2, SRC, SYK, and YES kinases were analyzed, and two approaches were taken to discover sequences with universal kinase coverage. The first in silico library of potential sequences was generated incorporating amino acid residues that were universally preferred. The second was generated using both commonly preferred and neutral amino acids. We then synthesized an initial set of nine peptides predicted to have broad specificity and evaluated them in kinase assays. The screen identified two hit sequences that were phosphorylated by the entire panel of PTKs tested. Further kinetic analysis confirmed these peptides to act generally as robust PTK substrates. The amino acid preferences at positions neighboring the tyrosine phosphorylation site for each of the kinases in the panel were assembled and used as previously described (Tables S1−14).14 A certain amino acid residue was considered “favored” if it was found to be present at a given position >2 standard deviations above the mean frequency of what would be expected given the “background” of potential tyrosinecentered sequences in a given substrate protein and any known interacting partners. We designated residues to be “neutral” to “slightly favored” as those exhibiting frequency values between −1 and 2 standard deviations from the mean, and “disfavored” as those observed at frequencies ≤1 standard deviation from the mean. In the case of HCK and PYK2, additional data from positional scanning peptide library screens were included like they were previously.14 We used these data in two types of approaches, one “unbiased” (library 1) and one “biased” (library 2), by choosing certain options from the potential amino acids represented in the motif. In the “unbiased” approach, the amino acids that were first, second, and third most favored by the kinases, as well the single most commonly observed neutral amino acid, were selected to generate the first sequence library. In the “biased” approach, the second library was generated by manual selection of amino acids that were either favored or neutral, and that were most commonly represented in the motifs across all kinases in the panel. These parallel approaches were used to provide sequences that could balance the dual needs of substrate efficiency and universal tolerance. A summary of the preferred and neutral amino acids used to construct these libraries is included in Figure S1. The sequences in the two libraries were obtained by generating every permutation of the particular amino acid at each position. The first library generated 663552 potential sequences, while the second library generated 2500 potential sequences. These sequences were then subjected to the Screener tool of the KINATEST-ID pipeline to score and rank their predicted likelihood as substrates for the set of PTKs in the panel as previously described.14 This tool calculates the sum of the PSSM-derived scores for each sequence relative to each kinase in the panel. It then ranks the sequences according to this score to predict potential “selectivity”, in which a higher score means that more kinases are predicted to phosphorylate the sequence and a lower score means it may be more selective

Table 1. List of the Nine Peptide Sequences Selected for Synthesis and Initial Evaluation in a Kinase Panel Activity Screena peptide

sequence

Screener sum score

1 2 3 4 5 6 7 8 9

EDDEYVTPE EDPIYVTLE DEDIYGTPE DEPIYDTPE DEAIYATVA DEPIYDTVE EDDVYDSVP EDDEYISPE EDDEYATPE

805.8 933.4 1035.9 1067.2 695.2 1055.5 987.1 676.5 667.1

a

Each of these was synthesized with an C-terminal -GGKbiotinGG tag for affinity capture purposes.

This pilot library was then subjected to an initial screen to test their phosphorylation by a panel of 15 PTKs that included 12 nonreceptor TKs (representing all 10 NRTK families) and three receptor TKs (representing three RTK families). Pilot peptides were assayed alongside several previously reported substrates that were designed to be specific to individual kinases.15 Conversion to product was monitored via incorporation of phosphate from γ-33P radiolabeled adenosine triphosphate (ATP). These intensities were normalized to background and visualized as a heat map in Figure 1. Peptides 2, 5, and 6 were phosphorylated by every PTK tested. By contrast, as previously reported, peptides that had been designed to specifically target individual kinases had more restricted reactivity. Peptides 2 and 5 had the highest average rate of phosphorylation across the entire panel of kinases (Figure 1B), in most cases as high as or even higher than that observed for the positive control peptides for a given kinase, and were chosen for subsequent characterization. To more fully characterize their properties as substrates for the PTKs tested, we performed follow-up kinetic analyses with B

DOI: 10.1021/acs.biochem.8b00044 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry

determined. While we found substantial variability between replicates for some kinases, relative rates of phosphorylation of the two peptides were highly consistent. For a given kinase, specificity constants for the two peptides were generally within 2-fold of each other. However, some kinases had a more substantial preference for either peptide 2 (EGFR) or peptide 5 (ABL and BTK). All nonreceptor TKs had kcat/Km values of ≥960 M−1 s−1 for at least one substrate. Kinetic parameters were generally less favorable than those reported for individually optimized consensus substrates,15 which was anticipated as generating peptides with broad reactivity necessitates the presence of suboptimal residues at some positions. Somewhat surprisingly, we found that for ABL and SRC, the catalytic parameters for peptides 2 and 5 rivaled those determined with consensus peptides. These results suggest that these two kinases are likely to be flexible in their primary sequence requirements. The lower activity for the three receptor TKs tested (FLT3, EGFR, and ALK) in comparison with that with the NRTKs may reflect the basal activity of these kinases, each of which has low activity in the absence of cancerassociated activating mutations.15,16 In conclusion, we used a modified workflow of the KINATEST-ID pipeline to design, completely in silico, a set of potential universal PTK substrate sequence motifs. It should be noted that none of the sequences in library 1 or 2 could be predicted to be comprehensively universal substrates based on the Screener filtering tool. The capabilities of Screener are limited by the input data sets for the PSSMs, which for understudied kinases could not be accurately modeled because of the small number (e.g., 15−20) of verified substrates identified in the literature. Nevertheless, the pilot collection of nine peptides yielded two hit sequences, a hit rate far better than those of typically achieved with randomized peptide libraries. These two hits, peptides 2 and 5, were first identified in an initial kinase activity screen and then further characterized via the determination of their kinetic parameters with a panel of seven PTKs representing a number of tyrosine kinase families. Both peptides were successful substrates for all kinases tested. Among the kinases we examined, peptide 2 was generally preferred over peptide 5 for RTKs, while the reverse preference was seen for NRTKs, suggesting differential utility for distinct classes of PTKs. We anticipate that these universal PTK sequences will be valuable tools for assaying enzymatic activity and screening for inhibitors in cases in which suitable substrates have not been previously described. Furthermore, because these sequences have such broad activity, they could be used generally as quality control tools for recombinant tyrosine kinase preparations and might enable more standardized

Figure 1. (A) Initial screen of the pilot library to test substrate suitability against a panel of 15 PTKs. (B) Peptides 2 and 5 had the highest phosphorylation averaged across the kinase panel (each point is an individual kinase; lines show the mean and 95% confidence interval) and were therefore designated universal substrate hits. Consensus peptides reported previously by the Turk lab for the respective kinases (labeled ACK, LYN, BMX, SYK, CSK, and ABL in both panels) were used as controls (sequences listed in the Supporting Information).

seven of the kinases from the panel (nonreceptor TKs BTK, SYK, ABL, and SRC and receptor TKs ALK, EGFR, and FLT3). Results for each kinase−peptide combination are summarized in Table 2. In most cases, Km values were too high (>150 μM) for individual parameters to be accurately determined, and for the remaining kinase−substrate pairs, the Km was in the double-digit micromolar range. As screening substrates are typically used at low micromolar concentrations, we used the kcat/Km value (the rate constant at low subsaturating peptide concentrations, also called the specificity constant) to compare substrates, and we extrapolated from this value a lower limit for the kcat value where it could not be

Table 2. Michaelis−Menten Kinetic Analysis for the Top Two Substratesa peptide 2 FLT3 ALK EGFR Abl BTK Syk Src

Km (μM)

kcat (s−1)

41 ± 5 >200 >200 >200 64 ± 11 >200 >200

0.013 ± 0.004 >0.02 >0.005 >4 0.061 ± 0.019 >2 × 10−7 >40

peptide 5 kcat/Km (s−1 M−1)

Km (μM)

kcat (s−1)

± ± ± ± ± ± ±

63.5 ± 0.2 >200 >200 101 ± 19 84 ± 16 >200 >200

0.014 ± 0.004 >0.02 >0.001 16.5 ± 3.5 0.26 ± 0.08 >0.1 >2

304 95 26 20100 970 1080 17500

63 19 5 2000 240 380 3800

2/5 kcat/Km (s−1 M−1) 215 79 5 160400 3170 560 9500

± ± ± ± ± ± ±

54 16 1 5700 840 120 2800

kcat/Km ratio 1.43 1.23 5.8 0.13 0.31 1.8 2.0

± ± ± ± ± ± ±

0.07 0.02 0.2 0.01 0.01 0.2 0.3

a Mean ± standard error of the mean of fits across experiment for all values; the kcat/KM for data which could not be fitted was derived from the slope of the linear portion of the titration.

C

DOI: 10.1021/acs.biochem.8b00044 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry benchmarking of sample quality and functional stability across kinases and preparations. Protein Expression and Purification. The mammalian expression vector for producing the intracellular portion of human FLT3 fused to GST (GST-FLT3564−993) was generated by amplifying the encoding sequence via polymerase chain reaction and subcloning into the BamHI and NotI sites of pEBG-2T. Mammalian expression vectors to produce BTK, SRC, and SYK as GST fusion proteins were previously described.15 GST-tagged kinases were expressed in HEK293T cells and purified by affinity chromatography using glutathione Sepharose 4B as described previously.15 His6-EGFR672−998 and His6-ALK1090−1416 were expressed and purified from Sf9 insect cells as described previously.16,17 Human ABL was purchased from Life Technologies. Universal Kinase Substrate Screen. Biotinylated peptides (10 μM final concentration) were arrayed in a 384-well plate in 18 μL of reaction buffer [50 mM HEPES (pH 7.5), 150 mM NaCl, 10 mM MgCl2, 1 mM DTT, and 0.1% (v/v) Tween 20]. Reactions were initiated by adding 2 μL of kinase (to final concentrations of 3−9 ng/μL) and [γ-33P]ATP (to a final concentration of 50 μM, 0.025 μCi/μL) in reaction buffer. Plates were sealed and incubated at 30 °C for 30 min, and then 2 μL aliquots were transferred to a streptavidin membrane (Promega SAM2 biotin capture membrane) using a pin tool. Membranes were washed as described previously,12 dried, and exposed to a phosphor screen. Incorporation of radiolabel was quantified by phosphorimaging using QuantityOne software (Bio-Rad). After background correction, data were normalized such that the peptide with the strongest signal was 100%. Normalized data from at least two separate experiments were averaged and converted to heat maps using Excel (Microsoft). Kinetic Analysis of Substrate Peptides. For kinetic experiments, versions of the 2 and 5 peptides were synthesized in which Lys residues were substituted for the two C-terminal glycine residues. Reactions were set up as described above, with varying concentrations of peptides (1, 2.5, 5, 7.5, 15, 25, 50, 75, 100, and 150 μM). Final kinase concentrations were as follows: 0.3 ng/μL for ABL, BTK, SRC, and SYK; 1.5 ng/μL for FLT3; and 2.7 ng/μL for ALK and EGFR. At various incubation times (10, 20, and 30 min), 2 μL aliquots were transferred using a pin tool to P81 phosphocellulose paper (Reaction Biology), which was washed three times in 75 mM H3PO4, briefly washed in acetone, and allowed to dry before being exposed to a phosphor screen. Incorporation of the radiolabel into peptides was quantified as described above. After background correction, initial rates of reaction were determined by linear regression and fit to the Michaelis−Menten equation using Prism (GraphPad). For kinase−peptide pairs that had Km values that were too high to accurately fit the data to the Michaelis− Menten equation, kcat/Km values were calculated from the slope of the linear portion of the Michaelis−Menten curve ([S] ≤ 75 μM). Reported kinetic parameters are the average of three separate experiments except for FLT3 (two experiments).

■

■

Scored and ranked sequence library 2 (XLSX) Peptide synthetic methods and characterization data (PDF)

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Laura J. Marholz: 0000-0002-9269-581X Benjamin E. Turk: 0000-0001-9275-4069 Author Contributions

L.J.M. and N.A.Z. contributed equally to this work. Funding

This work was supported by the National Institutes of Health (Grants R01CA182543, R33CA183671, and R01GM104047). L.J.M. was supported by a University of Minnesota Cancer Biology Training Grant (T32 CA009138) and a fellowship from the UMN Physical Sciences in Oncology Center grant (U54CA210190). N.A.Z. was supported by a Yale University School of Medicine Cancer Biology Training Grant (T32 CA1932000). Notes

The authors declare the following competing financial interest(s): Dr. Laurie Parker owns equity in and serves on the Scientific Advisory Board for KinaSense, LLC. The University of Minnesota and Purdue University review and manage this relationship in accordance with their conflict of interest policies.

■

ACKNOWLEDGMENTS The authors thank Jin Park in Mark Lemmon’s laboratory (Yale University) for generously providing EGFR and ALK for use in these studies. The authors also acknowledge Wei Cui (Purdue University) for his contributions to the design and synthesis of the pilot library.

■

REFERENCES

(1) Gschwind, A., Fischer, O. M., and Ullrich, A. (2004) The Discovery of Receptor Tyrosine Kinases: Targets for Cancer Therapy. Nat. Rev. Cancer 4, 361−370. (2) Arora, A., and Scholar, E. M. (2005) Role of Tyrosine Kinase Inhibitors in Cancer Therapy. J. Pharmacol. Exp. Ther. 315, 971−979. (3) Robinson, D. R., Wu, Y. M., and Lin, S. F. (2000) The Protein Tyrosine Kinase Family of the Human Genome. Oncogene 19, 5548− 5557. (4) Hayashi, H., Yamamoto, K., Yonekawa, H., and Morisawa, M. (1987) Involvement of Tyrosine Protein Kinase in the Initiation of Flagellar Movement in Rainbow Trout Spermatozoa. J. Biol. Chem. 262, 16692−16698. (5) Ali, N., Halfter, U., and Chua, N. H. (1994) Cloning and Biochemical Characterization of a Plant Protein Kinase that Phosphorylates Serine, Threonine, and Tyrosine. J. Biol. Chem. 269, 31626−31629. (6) Rębas, E., Lachowicz, L., Mussur, M., and Szkudlarek, J. (2001) The Activity of Protein Tyrosine Kinases of Rat Heart After Ischemia and Reperfusion. Med. Sci. Monit. 7, 884−888. (7) Varkondi, E., Schäfer, E., Bökönyi, G., Gyökeres, T., Orfi, L., Petak, I., Pap, A., Szokoloczi, O., Keri, G., and Schwab, R. (2005) Comparison of ELISA-Based Tyrosine Kinase Assays for Screenig EGFR Inhibitors. J. Recept. Signal Transduction Res. 25, 45−56. (8) Blouin, J., Roby, P., Arcand, M., Beaudet, L., and Lipari, F. (2011) Catalyltic Specificity of Human Protein Tyrosine Kinases Revealed by Peptide Substrate Profiling. Curr. Chem. Genomics 5, 115−121.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biochem.8b00044. Kinase Substrate Informatics sheets (ZIP) Scored and ranked sequence library 1 (XLSX) D

DOI: 10.1021/acs.biochem.8b00044 Biochemistry XXXX, XXX, XXX−XXX

Communication

Biochemistry (9) Rijksen, G., Adriaansen-Slot, S. S., and Staal, G. E. (1996) An Enzyme-Linked Immunosorbent Assay for the Determination of SrcFamily Tyrosine Kinase Activity in Breast Cancer. Breast Cancer Res. Treat. 39, 139−145. (10) Kobe, B., Kampmann, T., Forwood, J. K., Listwan, P., and Brinkworth, R. I. (2005) Substrate Specificity of Protein Kinases and Computational Prediction of Substrates. Biochim. Biophys. Acta, Proteins Proteomics 1754, 200−209. (11) Miller, M. L., Jensen, L. J., Diella, F., Jørgensen, C., Tinti, M., Li, L., Hsiung, M., Parker, S. A., Bordeaux, J., Sicheritz-Ponten, T., Olhovsky, M., Pasculescu, A., Alexander, J., Knapp, S., Blom, N., Bork, P., Li, S., Cesareni, G., Pawson, T., Turk, B. E., Yaffe, M. B., Brunak, S., and Linding, R. (2008) Linear Motif Atlas for PhosphorylationDependent Signaling. Sci. Signaling 1, ra2. (12) Mok, J., Kim, P. M., Lam, H. Y. K., Piccirillo, S., Zhou, X., Jeschke, G. R., Sheridan, D. L., Parker, S. A., Desai, V., Jwa, M., Cameroni, E., Niu, H., Good, M., Remenyi, A., Ma, J. L., Sheu, Y. J., Sassi, H. E., Sopko, R., Chan, C. S., De Virgilio, C., Hollingsworth, N. M., Lim, W. A., Stern, D. F., Stillman, B., Andrews, B. J., Gerstein, M. B., Snyder, M., and Turk, B. E. (2010) Deciphering Protein Kinase Specificity Through Large-Scale Analysis of Yeast Phosphorylation Site Motifs. Sci. Signaling 3, ra12. (13) Saunders, N. F. W., Brinkworth, R. I., Huber, T., Kemp, B. E., and Kobe, B. (2008) Predikin and PredikinDB: A Computational Framework for the Prediction of Protein Kinase Peptide Specificity and an Associated Database of Phosphorylation Sites. BMC Bioinf. 9, 245. (14) Lipchik, A. M., Perez, M., Bolton, S., Dumrongprechachan, V., Ouellette, S. B., Cui, W., and Parker, L. L. (2015) KINATEST-ID: A Pipeline to Develop Phosphorylation-Dependent Terbium Sensitizing Kinase Assays. J. Am. Chem. Soc. 137, 2484−2494. (15) Deng, Y., Alicea-Velázquez, N. L., Bannwarth, L., Lehtonen, S. I., Boggon, T. J., Cheng, H.-C., Hytönen, V. P., and Turk, B. E. (2014) Global Analysis of Human Nonreceptor Tyrosine Kinase Specificity Using High-Density Peptide Microarrays. J. Proteome Res. 13, 4339− 4346. (16) Park, J. H., Liu, Y., Lemmon, M. A., and Radhakrishnan, R. (2012) Erlotinib Binds Both Inactive and Active Conformations of the EGFR Tyrosine Kinase Domain. Biochem. J. 448, 417−423. (17) Bresler, S. C., Wood, A. C., Haglund, E. A., Courtright, J., Belcastro, L. T., Plegaria, J. S., Cole, K., Toporovskaya, Y., Zhao, H., Carpenter, E. L., Christensen, J. G., Maris, J. M., Lemmon, M. A., and Mossé, Y. P. (2011) Differential Inhibitor Sensitivity of Anaplastic Lymphona Kinase Variants Found in Neuroblastoma. Sci. Transl. Med. 3, 108ra114.

E

DOI: 10.1021/acs.biochem.8b00044 Biochemistry XXXX, XXX, XXX−XXX